Provider Guides

August 16, 2026 · View on GitHub

This index collects provider-specific guides for configuring VT Code with different LLM backends.

Provider whitelisting

Use providers_whitelist in vtcode.toml to restrict which providers VT Code may access. This is a governance control for environments where only approved inference endpoints should be reachable — for example, a corporate gateway or an air-gapped setup.

# Allow only corporate gateways + Gemini
providers_whitelist = ["opencode-zen", "opencode-go", "gemini"]

When providers_whitelist is non-empty:

  • The /model picker shows only the listed providers.
  • The first-run wizard offers only the listed providers.
  • The startup validator rejects agent.provider values not in the list.
  • Saving a model selection that falls outside the list is blocked.

When providers_whitelist is empty (the default), all built-in providers and [[custom_providers]] entries are available.

Whitelist entries may be a built-in provider key or a name from [[custom_providers]]. Matching is case-insensitive.

See the Configuration guide for full details.

Custom providers

Use [[custom_providers]] for an OpenAI-compatible endpoint that does not have a dedicated built-in provider. A custom provider can represent a private gateway, an aggregator, or an internal inference service.

Basic configuration

A typical entry includes a stable name, a human-friendly display_name, a base_url, and a default model. Set api_key_env when the endpoint requires an API key. Use models to expose several known model IDs in the model picker.

[[custom_providers]]
name = "mycorp"
display_name = "MyCorp"
base_url = "https://llm.corp.example/v1"
api_key_env = "MYCORP_API_KEY"
model = "gpt-5-mini"
models = ["gpt-5-mini", "gpt-5.4"]
context_window = 256000   # optional; defaults to 128000 tokens

Set the corresponding environment variable before launching VT Code:

export MYCORP_API_KEY="..."

Secure credentials are scoped by the custom provider name and api_key_env. They are not shared with another configured endpoint that happens to reuse the same environment variable.

Capability settings

context_window declares the provider capability in tokens. It controls the context size shown in the UI, automatic compaction, and preflight token checks. When omitted, the provider uses the default context window. The separate context.max_context_tokens setting can impose a lower session budget.

Use api_format when the endpoint needs an explicit request shape. Accepted values are auto, openai-chat, openai-responses, and anthropic-messages. Omitting the field preserves legacy autodetection. An explicit value is honored and does not silently fall back to another format.

Provider-level capability defaults are useful when the endpoint does not return model metadata. Supported defaults include:

  • supports_tools
  • supports_reasoning
  • supports_reasoning_effort
  • supports_vision
  • supports_structured_output
  • supports_parallel_tool_calls
  • supports_context_caching
  • supports_responses_compaction
  • supports_context_edits

Model profiles

Use sparse profiles for model-specific overrides:

[custom_providers.profiles."gpt-5.4"]
api_format = "openai-responses"
context_window = 131072
supports_tools = true
supports_vision = false
supports_structured_output = true
supports_parallel_tool_calls = true

Profiles apply only to an existing model identifier. They do not add models to the picker. Use model or models on the provider entry to control model availability.

When VT Code resolves a model's runtime shape, values are applied in this order, from highest to lowest priority:

  1. The matching per-model profile
  2. Provider-level capability defaults
  3. Metadata returned by the provider
  4. Conservative built-in defaults

An explicit boolean false is honored at every level. Omitting api_format preserves autodetection, while an explicit value selects that API shape.

Validate the configuration

List built-in and custom providers, then open the provider configuration flow:

vtcode models list
vtcode models config

Run a simple request after configuration:

vtcode ask "Summarize this repository"

Worked examples: Atlas Cloud and OmniRoute. See the Configuration guide for the complete field reference and precedence rules.

Google Gemini

  • Official docs: Gemini API models · Gemini 3.7 Flash
  • Provider key: gemini (env: GEMINI_API_KEY or GOOGLE_API_KEY)
  • Default model: gemini-3-flash-preview
  • Curated models:
    • gemini-3.7-flash — latest flash model, 1M context, tunable thinking (low/medium/high)
    • gemini-3.6-flash — flash model with improved reasoning and efficiency, 1M context
    • gemini-3.5-flash-lite — cost-optimized lightweight flash model, 1M context
  • Features: Streaming, tool calls, structured output, image/video/audio input, context caching, code execution, and configurable reasoning effort
  • Configuration details are covered in the main Getting Started guide.
  • Models and constants are defined in crates/codegen/vtcode-config/src/constants/models/google.rs.

OpenAI

  • Official docs:
  • Follow the Getting Started guide for API key setup.
  • See crates/codegen/vtcode-config/src/constants/models/openai.rs for the latest supported models.
  • Authentication methods (in priority order):
    1. ChatGPT subscription OAuth — vtcode login openai or /login openai. No API key needed. VT Code performs an in-process PKCE browser login with full auto-refresh. The Codex CLI is not required. By default, VT Code reuses Codex's public OAuth client identity as an unofficial compatibility mechanism (OpenAI has not documented or guaranteed third-party reuse, and a public client ID is not authorization to reuse another tool's registration). Organizations with their own OpenAI-issued client pair can override via VTCODE_OPENAI_OAUTH_CLIENT_ID / VTCODE_OPENAI_OAUTH_ORIGINATOR (both must be set together).
    2. Codex auth.json fallback — if you have Codex CLI installed and authenticated (codex login), VT Code automatically detects ~/.codex/auth.json and uses it at runtime when no VT Code-managed session is stored. Validate with vtcode login openai --from-codex.
    3. API key — vtcode secret add openai or set OPENAI_API_KEY. Use /secret to manage stored keys.
  • Login/logout commands:
    • CLI: vtcode login openai, vtcode login openai --from-codex, vtcode logout openai
    • TUI: /login openai, /logout openai, /auth
  • Logout semantics: vtcode logout openai (or /logout openai) clears VT Code's managed session only. If Codex's auth.json exists, VT Code will continue using it as a fallback until you run codex logout.
  • See the OAuth authentication guide for full details.
  • VT Code's default OpenAI profile is gpt-5.4 with reasoning_effort = "none" and verbosity = "medium"; raise reasoning only when the task shape justifies the extra latency.
  • VT Code applies a compact GPT-5.4 prompt contract rather than a verbatim cookbook prompt: compact outputs, low-risk follow-through, dependency-aware tool use, completeness checks, verification, and conditional grounding/citation rules.
  • Deprecated models (gpt-5, gpt-5-mini, gpt-5-nano, o3, o4-mini, gpt-5-codex, gpt-5.1-codex, etc.) are removed from the model picker but retained in routing constants for backward compatibility with existing configs.
  • File inputs are supported for native OpenAI Responses API requests through input_file parts.
  • Supported file input fields in VT Code message parts: file_id, file_data, file_url, filename.
  • file_url is Responses API only; VT Code rejects file_url when a request uses Chat Completions.
  • VT Code only upgrades local non-image file refs such as @report.pdf and @"Quarterly Deck.pptx" into structured file attachments for native OpenAI Responses sessions on api.openai.com.
  • Remote external document URLs such as @https://example.com/letter.pdf are also only elevated to structured file_url inputs for native OpenAI Responses sessions.
  • ChatGPT subscription sessions, OpenAI-compatible endpoints, and other providers keep non-image @file refs as plain text plus file-reference metadata so the agent can resolve the path and read the file with tools.
  • Raw image paths still use the existing multimodal image path flow. Non-image files require explicit @... references.
  • Official OpenAI Responses replays now preserve assistant phase metadata for replayed assistant history (commentary for preambles/progress updates, final_answer for completed answers) when the target GPT model supports it. VT Code does not send this field to Chat Completions, tool/user items, or non-native OpenAI-compatible endpoints.
  • OpenAI Responses hosted tools currently map through ToolDefinition for web_search, file_search, hosted tool_search, and remote mcp, with hosted config passed through directly on each tool entry.
  • OpenAI hosted shell mounts are configured through provider.openai.hosted_shell in vtcode.toml.
  • Hosted shell skill mounts support both skill_reference and inline bundle entries; VT Code forwards them to OpenAI but does not upload/create hosted skills in this path.
  • This hosted-shell workflow is separate from VT Code's local SKILL.md filesystem skills.
  • For large corpora, prefer File Search/Retrieval instead of sending full files inline.
  • For spreadsheet-heavy analysis, use Hosted Shell workflows instead of large inline sheet prompts.

Anthropic

  • Provider key: anthropic (env: ANTHROPIC_API_KEY)
  • Default model: claude-sonnet-5
  • Curated models: claude-sonnet-5, claude-fable-5, claude-mythos-5, claude-opus-5, claude-opus-4-8, claude-sonnet-4-6, and claude-haiku-4-5
  • Key management and defaults mirror the Gemini/OpenAI flow in Getting Started.
  • Supported model IDs live in crates/codegen/vtcode-config/src/constants/models/anthropic.rs.

DeepSeek

  • Provider key: deepseek
  • Authentication: DEEPSEEK_API_KEY environment variable
  • Base URL: https://api.deepseek.com/v1, override with DEEPSEEK_BASE_URL
  • Default model: deepseek-v4-pro
  • Curated models:
    • deepseek-v4-pro — high-performance reasoning model with advanced thinking capabilities
    • deepseek-v4-flash — official release with significantly enhanced agent capabilities for coding and tool use
  • Features: Streaming, tool calls, structured output, and reasoning support

xAI (Grok)

  • Official docs: xAI Docs · Models
  • Provider key: xai
  • Authentication: XAI_API_KEY environment variable
  • Setup: Set XAI_API_KEY from the xAI Console, then configure provider = "xai" in vtcode.toml
  • Default model: grok-4.6
  • Curated models:
    • grok-4.6 — flagship reasoning model, 500k context, reasoning_effort (low/medium/high/xhigh)
    • grok-4.5 — previous flagship reasoning model, 500k context
    • grok-4.3 — balanced general-purpose model, 1M context
    • grok-build-0.1 — fast coding model for agentic software engineering, 256k context
  • Features: Streaming, tool calls, structured output, image input, configurable reasoning effort, and 500k-token context

Meta AI

  • Guide: Meta AI Integration
  • Official docs: LLM documentation · Models
  • Provider key: meta
  • Authentication: META_API_KEY or Meta's documented MODEL_API_KEY
  • Base URL: https://api.meta.ai/v1, override with META_BASE_URL
  • Default model: muse-spark-1.2
  • Curated models: Muse Spark 1.2, Muse Spark 1.1, and the opt-in Muse Spark 1.2 Contributor tier
  • Features: Streaming, tool calls, structured output, multimodal input, reasoning effort, and 1M-token context

NVIDIA NIM

  • Guide: NVIDIA NIM Integration
  • Official docs: NVIDIA API catalog · NIM API reference
  • Provider key: nvidia
  • Authentication: NVIDIA_API_KEY (Bearer token)
  • Base URL: https://integrate.api.nvidia.com/v1, override with NVIDIA_BASE_URL
  • Default model: nvidia/nemotron-3-ultra-550b-a55b
  • Curated picker models: Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, GLM-5.2, and DeepSeek V4 Flash
  • Features: Streaming, tool calls, structured output, 1M-token context, and NVIDIA reasoning-content extraction

Merge Gateway

  • Guide: Merge Gateway Integration · Quick Reference
  • Official docs: Merge Gateway API overview · Coding-agent integration
  • Provider key: merge-gateway
  • Authentication: MERGE_GATEWAY_API_KEY (Bearer token; create a key in the Merge dashboard)
  • Base URL: https://api-gateway.merge.dev/v1 (native Responses), override with MERGE_GATEWAY_BASE_URL; explicit /v1/openai selects legacy Chat Completions
  • Default model: default_routing
  • Curated picker models: openai/gpt-5.5, anthropic/claude-opus-5, google/gemini-3.6-flash, google/gemini-3.7-flash, deepseek/deepseek-v4-pro-0813, deepseek/deepseek-v4-flash-0731, xai/grok-4.6, qwen/qwen3.8-max, minimax/minimax-h3, moonshot/kimi-k3, thinkingmachines/inkling, meta/muse-spark-1.1, openai/gpt-5.6-luna, openai/gpt-5.6-sol, and openai/gpt-5.6-terra
  • Features: Native Responses, streaming, tool calling, structured outputs, authenticated paginated model catalog, cache-backed picker metadata, and arbitrary explicit Merge route IDs
  • Limitations: Reasoning controls remain route-specific and are not inferred generically; routing metadata and billed cost remain outside VT Code's normalized response fields. Explicit /v1/openai endpoints retain the legacy compatibility path.

GitHub Copilot

  • Guide: GitHub Copilot Managed Auth
  • Runtime dependency: copilot must be installed and runnable for login/logout
  • Optional fallback: gh is only used when VT Code probes an existing GitHub CLI auth session
  • Commands: vtcode login copilot, vtcode logout copilot, /login copilot, /logout copilot

OpenRouter Marketplace

  • Guide: OpenRouter Integration
  • Official docs:
  • Default model: xiaomi/mimo-v2.5-pro (VT Code's default). Xiaomi MiMo V2.5 and V2.5 Pro are also available.
  • For Meta Muse, prefer the official meta provider when direct Meta access is desired. OpenRouter's meta/... entries are separately namespaced marketplace routes.
  • Meta Muse models via OpenRouter: meta/muse-glimmer-30b and meta/muse-spark-1.2
  • Curated picker catalog: openrouter/meta/muse-glimmer-30b, openrouter/meta/muse-spark-1.2, openrouter/deepseek/deepseek-chat, openrouter/moonshotai/kimi-k3, openrouter/moonshotai/kimi-k2.6, openrouter/moonshotai/kimi-k2.7-code, openrouter/qwen/qwen3.7-max, openrouter/tencent/hy3-preview, openrouter/x-ai/grok-build-0.1, openrouter/x-ai/grok-4.6, openrouter/xiaomi/mimo-v2.5, openrouter/xiaomi/mimo-v2.5-pro, openrouter/poolside/laguna-m.1:free, openrouter/poolside/laguna-s-2.1:free, openrouter/google/gemini-3.5-flash-lite, openrouter/google/gemini-3.6-flash, openrouter/google/gemini-3.7-flash, and openrouter/qwen/qwen3.8-27b
  • Xiaomi MiMo models:
    • xiaomi/mimo-v2.5-pro — flagship agentic model, 1M context, reasoning + tool calls
    • xiaomi/mimo-v2.5 — omnimodal model, 1M context, reasoning + tool calls

Atlas Cloud

  • Guide: Atlas Cloud Integration
  • Official docs:
  • Integration mode: configure Atlas Cloud through VT Code's [[custom_providers]] support because the LLM endpoint is OpenAI-compatible.
  • Base URL: https://api.atlascloud.ai/v1
  • Recommended model: start with deepseek-ai/deepseek-v4-flash (DeepSeek's latest flash model, 1M context, $0.14/M input tokens).

OmniRoute

  • Guide: OmniRoute Integration
  • Integration mode: configure OmniRoute through [[custom_providers]] as an OpenAI-compatible gateway.
  • Local base URL: http://localhost:20128/v1
  • Default model: auto delegates model selection and fallback to OmniRoute.
  • Features: Chat Completions, optional Responses API routing for compatible models, streaming, and function tools through VT Code's shared OpenAI-compatible transport.

Xiaomi MiMo

  • Provider key: mimo
  • Docs: Xiaomi MiMo Platform
  • Pricing: Pay-as-you-go · Subscription · Quick Access
  • Setup: Set MIMO_API_KEY or use the MiMo provider in VT Code's configuration
  • Models:
    • mimo-v2.5-pro — flagship agentic model, 1M context, deep thinking
    • mimo-v2.5 — omnimodal model (text, image, audio, video), 1M context

Ollama Local & Cloud Models

  • Guide: Local Inference Servers (unified /local command)
  • Setup: Install and run Ollama locally (official install)
  • Configuration: Local usage needs no key; set OLLAMA_API_KEY to access Ollama Cloud
  • Default model: gpt-oss:20b (local); any locally available model works
  • Curated picker catalog: gpt-oss:20b, gemma4, plus cloud models deepseek-v4-flash:cloud, deepseek-v4-pro:cloud, nemotron-3-ultra:cloud, kimi-k3:cloud, minimax-m3:cloud, and glm-5.2:cloud (also gpt-oss:120b-cloud via the OpenAI OSS support)
  • Cloud models: Use IDs like gpt-oss:120b-cloud with OLLAMA_BASE_URL=https://ollama.com
  • Custom Models: Use the custom-ollama option in the model picker to enter any locally or cloud-available model ID
  • Base URL: Configurable via OLLAMA_BASE_URL environment variable (defaults to http://localhost:11434)
  • Features: Streaming, structured tool calling (including Ollama's web search tools), and thinking traces when reasoning_effort is enabled

LM Studio Local Server

  • Guide: LM Studio Provider Guide
  • Server: Enable the OpenAI-compatible Developer server in LM Studio (defaults to http://localhost:1234/v1)
  • Environment: Optional LMSTUDIO_API_KEY when auth is enabled; override host/port via LMSTUDIO_BASE_URL
  • Default model: lmstudio-community/openai-gpt-oss-20b (local inference)
  • Catalog: Also ships with lmstudio-community/meta-llama-3.1-8b-instruct and lmstudio-community/gemma-3-12b-it, plus any custom GGUF models you expose
  • Features: Streaming, tool calling, structured output, and reasoning effort passthrough via the shared OpenAI surface

llama.cpp Local Server

  • Guide: llama.cpp Provider Guide
  • Server: VT Code targets llama-server and defaults to http://localhost:8080/v1
  • Environment: LLAMACPP_BASE_URL overrides the endpoint; LLAMACPP_MODEL_PATH enables VT Code-managed startup
  • Managed startup: VT Code can launch llama-server -m /path/to/model.gguf --port ... when the endpoint is localhost and a GGUF path is configured
  • Starter catalog: gpt-oss-20b, gemma-4-26b-a4b, gemma-4-e4b, and step-3.5-flash
  • Features: Streaming, dynamic /v1/models discovery, local no-auth defaults, and OpenAI-compatible request handling
  • Provider key: evolink
  • Official docs: Evolink Docs
  • Base URL: https://direct.evolink.ai/v1
  • Auth: EVOLINK_API_KEY environment variable
  • Setup: Set EVOLINK_API_KEY from Evolink dashboard, then configure provider = "evolink" in vtcode.toml
  • Models:
    • evolink/gpt-5.2 (default)
    • evolink/gpt-5.5
    • evolink/deepseek-v4-pro
    • evolink/deepseek-v4-flash
    • evolink/doubao-seed-2.0-pro
    • evolink/gemini-3.1-pro-preview
    • evolink/gemini-3.5-flash
    • evolink/MiniMax-M3
    • evolink/claude-sonnet-4-6
    • evolink/claude-opus-4-8
    • evolink/claude-haiku-4-5-20251001
  • Features: OpenAI-compatible gateway exposing many upstream models behind one endpoint. Evolink serves models under bare upstream names (e.g. gpt-5.2) that collide with VT Code's first-class providers, so curated model IDs are namespaced as evolink/<model>. The provider strips the prefix before sending requests upstream.

Anthropic API Compatibility Server

VT Code provides compatibility with the Anthropic Messages API to help connect existing applications to VT Code, including tools like Claude Code.

  • Feature: Anthropic API compatibility server
  • Command: vtcode anthropic-api --port 11434
  • Endpoint: /v1/messages (mirrors Anthropic Messages API)
  • Environment variables:
    • ANTHROPIC_AUTH_TOKEN=ollama (required but ignored)
    • ANTHROPIC_BASE_URL=http://localhost:11434
    • ANTHROPIC_API_KEY=ollama (required but ignored)
  • Features: Streaming, tool calling, vision support, multi-turn conversations

Z.AI (ZAI)

  • Provider key: zai
  • Official docs: Z.AI Platform
  • Auth: ZAI_API_KEY environment variable
  • Setup: Set ZAI_API_KEY from Z.AI platform, then configure provider = "zai" in vtcode.toml
  • Models:
    • glm-5.3 — flagship coding model, 1M context, reasoning + tool calls
    • glm-5.2 — flagship model for long-horizon tasks, 1M context, reasoning + tool calls
    • glm-5.1 — next-gen foundation model, reasoning + tool calls
    • glm-4.7 — efficient model for general tasks
  • Default: glm-5.3
  • Features: Streaming, tool calling, reasoning effort support

Moonshot (Kimi)

  • Provider key: moonshot
  • Official docs: Moonshot Platform
  • Auth: MOONSHOT_API_KEY environment variable
  • Setup: Set MOONSHOT_API_KEY from Moonshot platform, then configure provider = "moonshot" in vtcode.toml
  • Models:
    • kimi-k3 — 2.8T parameter flagship with Delta Attention, native vision, 1M context
    • kimi-k2.7-code — most capable coding model with long-horizon coding breakthrough, 256K context
    • kimi-k2.6 — multimodal model for coding and UI/UX generation, 1M context
    • kimi-k2.5 — enhanced reasoning model
  • Default: kimi-k3
  • Features: Streaming, tool calling, reasoning support, multimodal input (text, image, video)

StepFun

  • Provider key: stepfun
  • Official docs: StepFun Platform
  • Auth: STEPFUN_API_KEY environment variable
  • Setup: Set STEPFUN_API_KEY from StepFun platform, then configure provider = "stepfun" in vtcode.toml
  • Models:
    • step-3.7-flash — efficient reasoning model based on sparse MoE architecture
  • Default: step-3.7-flash
  • Features: Streaming, tool calling, reasoning support

MiniMax

  • Provider key: minimax
  • Official docs: MiniMax Platform
  • Auth: MINIMAX_API_KEY environment variable
  • Setup: Set MINIMAX_API_KEY from MiniMax platform, then configure provider = "minimax" in vtcode.toml
  • Models:
    • MiniMax-M3 — frontier multimodal coding model, 1M context
    • MiniMax-M2.7 — recursive self-improvement with enhanced reasoning
    • MiniMax-M2.5 — efficient model for general tasks
  • Default: MiniMax-M3
  • Features: Streaming, tool calling

HuggingFace

  • Provider key: huggingface
  • Official docs: HuggingFace Inference API
  • Auth: HF_TOKEN environment variable
  • Base URL: https://router.huggingface.co/v1, override with HUGGINGFACE_BASE_URL
  • Setup: Set HF_TOKEN from HuggingFace settings, then configure provider = "huggingface" in vtcode.toml
  • Default model: openai/gpt-oss-120b:huggingface
  • Notable models: openai/gpt-oss-20b:huggingface, deepseek-ai/DeepSeek-R1, deepseek-ai/DeepSeek-V4-Pro:together, deepseek-ai/DeepSeek-V4-Pro:novita, zai-org/GLM-5.1:zai-org, zai-org/GLM-5.2:novita, moonshotai/Kimi-K3:together, moonshotai/Kimi-K2.6:novita, MiniMaxAI/MiniMax-M3:novita, MiniMaxAI/MiniMax-M2.7:novita, stepfun-ai/Step-3.5-Flash:featherless-ai
  • Features: Access to various models through HuggingFace's inference API, including models from OpenAI, DeepSeek, Z.AI, Moonshot, MiniMax, and other providers

Poolside

  • Provider key: poolside
  • Auth: POOLSIDE_API_KEY environment variable
  • Base URL: https://api.poolsi.de/openai/v1, override with POOLSIDE_BASE_URL
  • Default model: poolside/laguna-s-2.1
  • Curated picker models: poolside/laguna-s-2.1, poolside/laguna-m.1, poolside/laguna-xs.2
  • Setup: Set POOLSIDE_API_KEY from Poolside platform, then configure provider = "poolside" in vtcode.toml

Mistral

  • Provider key: mistral
  • Authentication: MISTRAL_API_KEY environment variable
  • Base URL: https://api.mistral.ai/v1, override with MISTRAL_BASE_URL
  • Default model: mistral-large-2512
  • Curated picker models: mistral-large-2512, mistral-medium-3-5, mistral-small-2603, mistral-medium-2508, codestral-2508
  • Features: Streaming, tool calls, structured output, and reasoning support

Qwen

  • Provider key: qwen
  • Authentication: QWEN_API_KEY (alternate DASHSCOPE_API_KEY)
  • Base URL: https://dashscope.aliyuncs.com/compatible-mode/v1, override with QWEN_BASE_URL
  • Default model: deepseek-v4-flash
  • Curated picker models: deepseek-v4-flash, deepseek-v4-pro, glm-5.1
  • Features: Streaming, tool calls, and reasoning support

OpenCode Zen

  • Provider key: opencode-zen
  • Authentication: OPENCODE_ZEN_API_KEY environment variable
  • Base URL: https://opencode.ai/zen/v1, override with OPENCODE_ZEN_BASE_URL
  • Default model: opencode/gpt-5.4
  • Curated picker models: opencode/gpt-5.4
  • Setup: Set OPENCODE_ZEN_API_KEY from the OpenCode Zen console, then configure provider = "opencode-zen" in vtcode.toml
  • Features: Curated pay-as-you-go gateway over flagship models

OpenCode Go

  • Provider key: opencode-go
  • Authentication: OPENCODE_GO_API_KEY environment variable
  • Base URL: https://opencode.ai/zen/go/v1, override with OPENCODE_GO_BASE_URL
  • Default model: opencode-go/glm-5.1
  • Curated picker models: opencode-go/glm-5.1, opencode-go/glm-5.2, opencode-go/kimi-k2.7-code, opencode-go/kimi-k2.6, opencode-go/mimo-v2.5-pro, opencode-go/mimo-v2.5, opencode-go/minimax-m3, opencode-go/minimax-m2.7, opencode-go/qwen3.7-max, opencode-go/qwen3.7-plus, opencode-go/qwen3.6-plus, opencode-go/deepseek-v4-pro, opencode-go/deepseek-v4-flash
  • Setup: Set OPENCODE_GO_API_KEY from the OpenCode Go console, then configure provider = "opencode-go" in vtcode.toml
  • Features: Subscription-based access to flagship open models for agentic coding

ℹ Additional provider-specific guides will be added as new integrations land in VT Code.